1

Application of the PC-SAFT Equation of State to Asphaltene Phase Behavior

P. David Ting,1,2 Doris L. Gonzalez,2 George J. Hirasaki,2 and Walter G. Chapman2

1 – Work performed while at RiceUniversity, Currently with Shell Global Solutions (US) Inc., Westhollow Technology Center, Houston, TX 77082

2 - Department of Chemical Engineering, Rice University, Houston, TX 77005

Abstract

A method to characterize crude oil including asphaltenes using the perturbed chain form of the

statistical associating fluid theory (PC-SAFT) is presented. The theory accurately predicts the

bubble point, density, and asphaltene precipitation onset for the oil. Examples show that the

theory predicts asphaltene instability due to changes in pressure, temperature, and fluid

composition. Further work demonstrates the effect of asphaltene polydispersity and resins on the

phase behavior of asphaltenes. The approach demonstrates that laboratory and field observations

of asphaltene phase behavior can be explained based only on molecular size and van der Waals

interactions.

I. Introduction

This chapter provides an application of the Statistical Associating Fluid Theory (SAFT)1-4

equation of state (EOS) to model the effects of pressure, temperature, and composition on the

phase behavior and stability of asphaltenes in crude oil. SAFT is a versatile molecular model

capable of predicting the effects of molecular shape, van der Waals forces, polar interactions,

and association on the thermodynamic properties and phase behavior of fluids. The approach we

2

have taken is to use the minimum number of (physically relevant) parameters possible to

describe phase behavior of asphaltenes by including only essential interactions.

Given the high degree of complexity of crude oil and its large variability in composition due, in

part, to differing source rock properties and migration history, one may question the relevance of

a molecular based model in light of our current understanding of asphaltenes. It is our belief that

the bulk phase behavior of asphaltenes can be accurately described if we can correctly account

for its major physical attributes and its interactions with other species in oil. There are many

advantages in using a predictive molecular based EOS model. For instance, the sensitivity of

asphaltene phase behavior to the effects of temperature, pressure, and asphaltene polydispersity

can be quickly and confidently modeled. An EOS framework can also be more readily

implemented into existing reservoir and thermal-hydraulic simulators used in industry. At the

other end of the scale, the EOS approach, when compared to more rigorous molecular

simulations, has the advantage of speed and ease of use. This comes, of course, at the expense of

detailed descriptions at the molecular level.

With the molecular basis of the SAFT approach, it is necessary to define the system to be

modeled. Our viewpoint is described and justified in the Section I of this chapter. In Section II,

we present in detail the characterization of an oil with SAFT including applications to asphaltene

phase behavior on reservoir depressurization and gas injection. Conclusions are provided in the

final section.

3

I.A. Asphaltene Properties & Field Observations

The currently accepted definition of asphaltenes is an operational one based on solubility (i.e.,

asphaltenes are insoluble in heptane or pentane and soluble in toluene). As such, it reveals little

about the structure of asphaltenes. Most researchers agree that asphaltenes are a polydisperse

mixture of molecules containing polynuclear aromatic, aliphatic, and alicyclic moieties with

small amounts of dispersed heteroelements such as oxygen, sulfur, nitrogen, vanadium, and

nickel. When compared to other crude oil components, asphaltenes are the heaviest fraction of a

distribution (in molecular weight as well as aromaticity) of compounds that include aromatics

and resins in the lower molecular weight sub-fractions. The accepted definition for asphaltenes

is, in essence, an arbitrarily divided sub-fraction of this distribution. Asphaltenes are more

aromatic (low H/C ratio) than most other oil fractions, are larger in molecular weight, and have

higher solubility parameter.5

The interesting phase behavior of asphaltenes in oil can be deduced by studying a few

representative examples of field experiences with asphaltene problems.6-11 These examples not

only elucidate asphaltene behavior typically seen in the field and in PVT laboratories but also

help us gain a better understanding of the type of molecular interactions between asphaltenes and

other oil species. From these experiences, it can be concluded that asphaltenes are usually stable

in heavier oils well above the bubble point pressure under reservoir conditions. Light oils with

little asphaltenes are the most susceptible to asphaltene problems. Field operators have reported

that asphaltenes are unstable over a range of pressures during reservoir depressurization. More

specifically, the effect of pressure on asphaltene phase separation is most pronounced for light

oil near the bubble point. Asphaltenes are stable at high pressure and at pressures well below the

4

bubble point, but, in many cases, they are unstable at pressures somewhat above and below the

bubble point. Far below the bubble point, asphaltenes tend to be stable since most of the

precipitants (methane, ethane, nitrogen, etc.) have escaped from the liquid. Compositional

changes such as oil blending or miscible flooding sometimes result in asphaltene precipitation.

Interestingly, temperature changes may result in either asphaltene precipitation or solubilization.

For instance, in the propane deasphalting process, asphaltenes become increasingly unstable with

temperature increase.12 However, for n-alkane (n-C5+) titrations, asphaltene stability improves

with increasing temperature.12 These examples highlight the need, by such disciplines as flow

assurance and production engineers, for a predictive model that is capable of explaining these

observations.

A distinction needs to be made between thermodynamic asphaltene stability (which is the focus

of this work) and asphaltene-related field problems. Just because asphaltenes may become

unstable during production does not necessarily mean that deposition will be encountered.

Factors such as (1) the properties of the asphaltene that precipitated including its “stickiness”, (2)

the amount of the asphaltene that precipitated, and (3) the flow pattern in the production system

(flowline, tree, riser, inside topside facility, etc.) all play a role in determining whether the

precipitated asphaltenes will result in field problems. However, the ability to accurately model

asphaltene as a polydisperse system will help in elucidating its deposition tendencies. Finally, it

is important to note that asphaltene from one region (source) may have very different properties

than asphaltene from another region.

5

I.B. The Two Views of Asphaltene Interactions

In the last 50+ years, two views to describe the phase behavior of asphaltenes have emerged. In

one view, which we will call the “molecular solution” approach, asphaltenes are treated as

molecules that are solubilized by the oil. Asphaltene precipitation is treated as liquid-liquid or

solid-liquid equilibria. The hypothesis in this framework has been that molecular size and van

der Waals interactions, which are related to molecular polarizability, dominate asphaltene phase

behavior in reservoir fluids; the more polarizable components (the resins and aromatics)

solubilize the asphaltenes and the less polarizable components (saturates) destabilize the

asphaltenes. Proponents of this view have used such approaches as Flory-Huggins theory13 or

equations of state to model asphaltene phase behavior.14-18 In this chapter we report on the

application of the SAFT equation of state to model asphaltene phase behavior.19,20 Although the

SAFT approach and its extensions can account for polar and association effects, we find that the

effects of molecular size and van der Waals attractions are sufficient to explain reported

observations of asphaltene phase behavior.

The application of Flory-Huggins-regular solution theory to describe asphaltene phase behavior

was first proposed by Hirschberg, et al.15 While the Flory-Huggins-regular solution based

approaches have been used with varying success to model asphaltene solubility with n-alkane

titrations under ambient conditions, it is difficult to extend the approach to model asphaltene

solubility under reservoir conditions. In a sense, the theory is not a “complete” equation of state;

it requires the molar volumes and solubility parameters under reservoir conditions. These values

must be obtained from an equation of state or estimated from empirical correlations. For

example, in the work of Chung, et al.,21 the Flory-Huggins-regular-solution model is combined

6

with the Peng-Robinson cubic equation to model asphaltene solubility in oil. And in the work of

Burke, et al.,22 the Flory-Huggins-regular-solution model parameters were obtained from the

Zudkevitch-Joffe-Redlich-Kwong equations. Some success in predicting asphaltene

precipitation at reservoir conditions from Flory-Huggins parameters fit to ambient titration data

has been reported using an extrapolation due to Wang and Buckley.14, 23

Another limitation of the Flory-Huggins model is its inability to predict certain classes of phase

behavior. As an example, the model requires temperature dependent binary interaction

parameters to show lower critical solution behavior, phase behavior known to occur in systems

with large size differences between molecules.

Another popular classical thermodynamics approach to model asphaltene behavior is to use cubic

equations of state. In the method proposed by Nghiem, et al.,24 the C31+ heavy end of crude oil is

first divided into nonprecipitating and precipitating sub-fractions. Different interaction

parameters (between these sub-fractions and light ends) are then assigned to reproduce

experimental results. In another example, Akbarzadeh, et al.16 modified the Soave-Redlich-

Kwong cubic equation by adding an additional aggregation size parameter to asphaltenes. The

cubic equations have relatively simple functional forms and are easy to implement into existing

reservoir simulators because cubic equations have been used extensively to describe the

thermodynamic behavior of reservoir fluids. However, the major shortcoming of cubic equations

of state is that they cannot describe the phase behavior of systems with large size disparities25

and that they cannot accurately describe liquid densities. Accurate modeling of liquid density is

essential for an equation of state to predict liquid-liquid equilibria over a range of conditions.

7

While the calculated densities can be modified using volume translation techniques, volume

translations do not affect phase equilibria calculations.

In our approach, asphaltene instability is modeled as liquid-liquid equilibria. We have chosen to

base our model on the SAFT equation of state because of its ability to accurately model fluid

densities as well as phase behavior for mixtures with substantial size asymmetry.20, 25 SAFT and

its extensions also explicitly account for association and polar interactions. The details of the

model are discussed in section II.

As noted above, the “molecular solution” approach is not the only approach used in literature to

describe asphaltene phase behavior in oil. In this other approach generically called the

“micellar” approach, the structural characteristics of asphaltenes and resins are emphasized -

asphaltenes are viewed to be stabilized by resins via polar-polar interactions. The basis of this

viewpoint is that asphaltenes and resins are the most polar fractions of crude oil because they

contain heteroatoms of various proportions. When resins are added, less asphaltenes will

precipitate. And when n-alkanes are added, asphaltenes will precipitate because of the dilution

of resins in the mixture. The argument is that resins stabilize asphaltenes in a similar way to

surfactants stabilizing a micro-emulsion in an oil/water mixture. Thermodynamic models that

take the micellar view include the solid-asphaltene colloidal model proposed by Leontaritis and

Mansoori,26 the reversible micellization model proposed by Victorov and Firoozabadi,27 and the

McMillan-Mayer-SAFT based theory proposed by Wu.12, 28, 29 Interestingly, the SAFT

framework is versatile enough that it has been used to develop models for both points of view.

8

I.C. Our View and Approach

The underlying hypothesis of our approach in modeling asphaltene phase behavior is that

molecular size and nonpolar van der Waals interactions dominate asphaltene phase behavior in

reservoir fluids. We find that the phase behavior described in field and laboratory experiences

are similar to those seen in oligomer and polymer systems. For instance, the pressure,

temperature, and compositional behavior of asphaltenes in reservoir fluid described in Section

1.1 is qualitatively similar to the behavior of polystyrene in a mixture of cyclohexane and CO2.30

In such systems, the phase behavior can be predictively modeled by only considering molecular

size and van der Waals interactions. Note that the existence of other types of intermolecular

interactions are not being discounted or trivialized in our framework; we are simply taking the

approach that the behavior of asphaltenes in crude oil systems can be described to a large extent

by accounting for molecular size and van der Waals interactions only.

Our hypothesis is supported by other evidence. For instance, an investigation of asphaltene

solubility in over 40 polar and nonpolar solvents by Wiehe31 shows that asphaltenes are soluble

in solvents with high field force solubility parameters (which is a measure of nonpolar, van der

Waals interaction strength) and insoluble in solvents with moderate and high complexing

solubility parameters (which is a measure of hydrogen bonding and polar interaction strengths).

While this observation is consistent with the idea that van der Waals interactions determine

asphaltene phase behavior, it does not readily fit in the micellar framework.

As another example, consider that some relatively nonpolar molecules of similar size and

structure can be either precipitants or solvents for asphaltenes. Toluene (C6H5CH3) is a good

9

solvent for asphaltenes while n-heptane (C7H16) is a precipitant. Similarly, both carbon disulfide

(CS2) and carbon dioxide (CO2) are weakly polar and of similar molecular structure, but CS2 is a

good solvent for asphaltenes while CO2 is a precipitant. In these examples, the more polarizable

molecules (such as toluene and carbon disulfide, molecules with the stronger van der Waals

interactions as shown by their solubility parameters) are the better solvents for asphaltenes.

In some situations, the role of polar or hydrogen bonding interactions is important. For example,

asphaltenes may aggregate on the water-oil interface and stabilize water emulsions.32 Also,

addition of a sufficiently large amount of alkyl-benzene derived amphiphiles (such as dodecyl

benzene sulfonic acid) can inhibit asphaltene aggregation.33 While polar interactions may play a

role, our hypothesis is that asphaltene phase behavior in the reservoir, such as on reservoir

depressurization, is shaped to a larger extent by the nonpolar interactions in the oil.

Recent structural investigations on asphaltene behavior34 suggest that multiple levels of

interactions may be occurring in fluids containing asphaltenes. In all except very good solvents

and under near-infinite dilution concentrations, asphaltenes exist as molecular aggregates (of

several units, each ~500-800 amu in size) and each aggregate behaves as if it is a single

molecule. Our approach at this point is to assume that asphaltenes have pre-aggregated in crude

oil systems and that the asphaltenes in our model would, in fact, exhibit characteristics of the

molecular aggregate so that the molecular weight in our model represents the MW of the

aggregate. This approach is justified since oil systems (and the model systems investigated in

this work) are neither infinitely dilute in asphaltenes nor very good solvents for asphaltenes.

10

II. Introduction to SAFT

We adopt the SAFT equation of state for our study of asphaltenes because of its demonstrated

ability to accurately describe and predict the effects of large molecular size differences and

association on phase behavior of complex mixtures.25, 30, 35, 36 This has been seen in numerous

applications of SAFT to polymer solutions and hydrogen bonding fluids. For example, SAFT

has become an important tool in predicting polymer phase behavior to prevent fouling in

polymer processing.37

SAFT was developed by Chapman, et al.1, 2, 3 based on extensions and simplifications of

Wertheim’s theory for associating fluids.38, 39, 40, 41 In SAFT, molecules are modeled as chains of

bonded spherical segments. As shown in Figure 1, SAFT determines the free energy of a fluid as

the sum of the free energy for a collection of spherical “segments” (from which molecules are

constructed) plus the change in free energy on “bonding” these spherical “segments” in a

prescribed manner to form the molecules of interest. This change in free energy can be

calculated from Wertheim’s theory. Finally, if the molecules have association sites, the change

in free energy due to these directional interactions are explicitly accounted for using Wertheim’s

theory.

A theory like SAFT that is based in statistical mechanics offers several advantages. The first

advantage is that each of the approximations made in the development of SAFT such as the chain

and association terms has been verified against molecular simulation results.1 In this way, the

range of applicability and the shortcomings of each term in the equation of state have been

assessed. A second advantage is that the SAFT parameters have physical meaning. For

11

example, a chain molecule is characterized by the diameter or volume of a segment, the number

of segments in the chain, and the segment-segment van der Waals attraction. These parameters

are, in general, fit to saturated liquid densities and vapor pressures for the pure components.

Since the equation of state parameters are physical, they behave systematically within a

homologous series. Furthermore, parameters for new systems can be estimated from those of

previously modeled systems. In this way, parameters for saturates, aromatics, and resins can

quickly be determined to model a crude oil.

Numerous forms of the SAFT equation of state have been proposed.1-4, 42-44 These forms differ

only in the segment term used to account for the van der Waals attraction between molecules; all

use the same chain and association terms as introduced in the original SAFT papers by Chapman,

et al.1-3, 42 Because each of these SAFT versions shares the same basic form of the equation of

state, they each give similar results. In this work, we report results using the Perturbed Chain

version of SAFT or PC-SAFT due to Gross and Sadowski.4 We expect qualitatively similar

results if another version of SAFT is used instead.

II.A PC-SAFT Pure Component Parameters

For each non-associating species in SAFT, the equation of state requires the specification of

three physical parameters: σ, the diameter of each molecular segment, m, the number of

segments in the molecule, and ε/k, the interaction energy (van der Waals attraction) between

each molecular segment. A list of PC-SAFT parameters for compounds of interest to this work

is given in Table 1.

12

An important feature of SAFT is that the fitted pure component parameters (σ, m, and ε/k)

behave in a systematic manner with molecular weight for different classes of compounds4, 45 (see

Figure 2). Furthermore, species with both aromatic and aliphatic characteristics have EOS

parameters that lie in between the aromatic and n-alkane correlations in a systematic manner

depending on their degree of aromaticity or aliphaticity. For instance, the parameters for

benzene derivatives and cyclo-alkanes approach the pure component parameters of the alkanes as

the degree of aliphaticity increases. These “well-behaved” correlations between the pure

component parameters and molecular weight have three implications. (1) SAFT parameters for

crude oil components or lumps of components (pseudo-components) can be estimated from their

average molecular weights. (2) SAFT parameters can be estimated for substances whose vapor

pressures and/or liquid densities are difficult to measure. In the case of asphaltenes, the

correlations of aliphatic and aromatic SAFT parameters with molecular weight provide the upper

and lower bounds of the asphaltene SAFT parameters. And (3) the “well-behaved” pure

component SAFT correlations imply that the EOS can easily be extended to model the effects of

polydispersity.

Note that the systematic behavior shown in Figure 2 has been observed for a number of SAFT

implementations (see previous section). Discussions from this point on will focus on the use of

PC-SAFT.

II.B PC-SAFT Characterization of a Recombined oil

To use the SAFT equation of state to model live oil systems, we need to be able to account for

the fluid’s (PVT) phase behavior using a minimum number of real components and “realistic”

13

pseudo-components. In this section, we will outline (1) a method to model live oil systems using

a small number of real components and pseudo-components and (2) a method to characterize

asphaltenes in the SAFT model. Both (1) and (2) take advantage of the systematic SAFT

parameter-molecular weight behavior discussed in Section II.A.

Consider the application of PC-SAFT to a recombined “live” oil (separator oil plus its

associated gas) for which the bubble point, GOR, stock tank oil density (API gravity), density

above saturation, and asphaltene precipitation onsets have been measured.47 To model the phase

behavior of the recombined oil, the oil is treated as a mixture of six pseudo-components (Figure

3). Three sub-fractions (methane, N2+CO2, and light n-alkanes) are used to describe the separator

gas and three sub-fractions (saturates, aromatics+polynuclear aromatics, and asphaltenes) are

used to describe the stock tank oil.

The relative amount of each sub-fraction given in Figure 3 can be calculated based on

composition data from gas chromatography (GC), SARA fractionation data, and GOR data. The

lumping of N2 with CO2 was appropriate in this case because very little CO2 was present. More

generally, we treat N2 and CO2 as separate components.

Note that the proposed lumping scheme introduces a small error in the description of the fluid

properties. This is because there are non-negligible amounts of light n-alkanes such as butane

and pentane partitioned in both the “light n-alkane” pseudo-components and the “saturates”

pseudo-component. However, the effect of this error on the fluid phase behavior was found to be

small.

14

II.B.1 Characterization of Separator Gas

The three PC-SAFT parameters (σ, m, and ε/k) for each pseudo-component are related to the

average molecular weight of that pseudo-component. Parameters for the pseudo-components can

be interpolated/extrapolated from correlations shown in Figure 2. For instance, the PC-SAFT

parameters for the “light n-alkane” pseudo-component lie on the n-alkane correlations in Figure

2 with an average molecular weight of 47. For a real component, PC-SAFT parameters are taken

from fits to the component’s saturated liquid densities and vapor pressures. Parameters for the

separator gas real- and pseudo-components are given in Table 2.

To accurately describe the van der Waals attraction between unlike pairs of molecules, the

energies of interaction are often modified in an equation of state via the use of binary interaction

parameters. The binary interaction parameters used in this work were fit to vapor-liquid

equilibria data, with the values listed in Table 3.

The PC-SAFT calculated density of 0.3708 g/cm3 obtained from the three pseudo-component

treatment of the separator gas compares well with the measured density of 0.3773 g/cm3 at 241.3

bar and 293K.

II.B.2 Characterization of Stock-Tank Oil

The stock-tank oil is treated as a mixture of 3 pseudo-components (saturates, aromatics+resins,

and asphaltenes). This particular division was set up partly to take advantage of the SARA

15

fractionation data. Unlike SARA, which breaks down an oil into saturates, aromatics, resins, and

asphaltenes, we grouped aromatics and resins into one single pseudo-component because they

form a part of the same distribution in molecular weight and aromaticity (that actually extends

out to the asphaltenes). The following assumptions are applied to the oil composition data to

determine the composition of each sub-fraction given in Figure 3:

• We assume that saturates can be distinguished from aromatics below the C10 carbon cut.

• Since 72.3% mass of the stock tank oil are made up of saturate components (from SARA

fractionation analysis of the example oil in this work), 72.3% mass of the materials from

the higher carbon cuts (the C10 cut and higher) are assumed to be saturate components.

The remaining 27.7% (mass) of the materials are assumed to be either aromatics+resins

(up to, and including the C29 cut) or aromatics+resins and asphaltenes (in the C30+ cut).

• The asphaltene fraction accounts for 2.5% mass (from SARA analysis) of the stock-tank

oil, and all asphaltenes are found in the C30+ sub-fraction.

• The molecular weight distribution (and thus the average molecular weight) of the C30+

saturates follows an exponential distribution in mole amount

• The components in the saturates sub-fraction are all normal alkanes.

Based on these assumptions, the average molecular weights were first determined and the PC-

SAFT parameters subsequently obtained (from correlations in terms of average molecular weight

in Figure 2) for the saturates pseudo-components. The parameters are presented in Table 2.

Binary interaction parameters are presented in Table 3 based on fits of vapor-liquid equilibria

16

data for saturates. SAFT is one of the few equations of state capable of accurately predicting

liquid-liquid equilibria based on binary interaction parameters fit to vapor-liquid equilibria data.

In the case of the aromatics+resins pseudo-component, the PC-SAFT parameters depend not

only on the average molecular weight but also on the average aromaticity of the sub-fraction.

This is because the aromatics+resins sub-fraction is a mixture of aromatics, polynuclear

aromatics (PNA), PNA derivatives, and resins. Since the PC-SAFT parameters for compounds

in each of these classes follow different parameter correlation curves, it is important to quantify

this “degree of aromaticity”. An aromaticity parameter was defined to interpolate between

parameters for PNA’s and those for aromatic derivatives. For a stock tank oil, the degree of

aromaticity was adjusted (1) to match the stock tank oil density and (2) so that a “model” n-C7

insoluble asphaltene is completely soluble in a fluid composed of only the aromatics+resins

fraction. Hence, the aromatics+resins pseduo-component possesses the minimum degree of

aromaticity necessary to dissolve asphaltenes. The binary interaction parameter between the

saturates and the aromatics+resins pseudo-components are set to –0.01 based on the optimal kij

for toluene-dodecane and toluene-hexadecane VLE data57 at 353K and 333K.

At this point, all PC-SAFT parameters have been determined except for the parameters for the

asphaltenes. We fit the PC-SAFT parameters for asphaltenes to data for asphaltene precipitation

on n-alkane titration of the stock tank oil (see Figure 4). For the oil illustrated in this work, the

refractive index of the mixture at the initial point of asphaltene precipitation (PRI) was measured

by Wang and Buckley.58 The PRI is related to the composition and density of the oil; it can also

be correlated to the solubility parameter of the oil. In measuring the PRI, the stock-tank oil was

17

initially mixed with an equal volume of alpha-methyl naphthalene to re-dissolve any precipitated

material. For a given oil/precipitant pair, the addition of asphaltene solvents to the oil shifts the

amount of precipitant needed to induce asphaltene precipitation but does not significantly alter

the refractive index at precipitation onset.59 By simulating these experiments using PC-SAFT,

we are able to fit the three PC-SAFT parameters for asphaltenes to reproduce the PRI behavior.

In these calculations, to minimize the number of adjustable parameters, the binary interaction

parameters between asphaltenes and the other stock-tank oil components (and alpha-methyl-

naphthalene) were set to zero. Using these asphaltene parameters PC-SAFT predicts a solubility

parameter for the asphaltene of 21.85 MPa0.5 that is consistent with values reported in the

literature.5, 15, 22

II.C Comparison of Results & Analysis of Asphaltene Behavior

The fit of asphaltene parameters to titration data is shown in Figure 4 in terms of the refractive

index of the mixture at the initial point of asphaltene precipitation (PRI). Although the asphaltene

parameters were fit to ambient condition alkane titration experiments (for an oil blended with an

equal volume of alpha-methyl naphthalene), our experience is that these same asphaltene

parameters are applicable for asphaltenes in uncontaminated oil at reservoir conditions. PC-

SAFT parameters for the asphaltene component are given in Table 2. This asphaltene has a PC-

SAFT calculated solubility parameter of 21.85 MPa0.5 and a molar volume of 1,437 cm3/mol.

Given that the density of asphaltenes is estimated to be between 1.13 g/cm3 and 1.20 g/cm3, the

SAFT calculated stock-tank oil asphaltene has an implied MW between 1,624 and 1,724.

Because the asphaltenes are assumed to be pre-aggregated in the oil, this reflects the MW for the

pre-aggregated asphaltenes.

18

A comparison of PC-SAFT-predicted and experimental densities for the one-phase recombined

oil with a gas-oil ratio of 152 m3/m3 (for the reservoir fluid) and a GOR of 212 m3/m3 (for the

reservoir fluid with additional separator gas added) is shown in Figure 5. Although PC-SAFT

slightly under-predicts the recombined oil densities, the predicted densities are in good

agreement with experimental values.

The PC-SAFT predicted asphaltene instability onset and bubble point pressures for the example

recombined oil at reservoir temperature (71.1°C) are plotted versus experimental results in

Figure 6 as a function of separator gas composition (corresponding to GORs’ of 152 m3/m3 and

212 m3/m3). PC-SAFT accurately predicts both the bubble points and the asphaltene instability

onset points for this oil in the composition range investigated. According to the equation of state

calculations, we would expect asphaltene precipitation problems to occur at higher pressure for

higher separator gas concentrations.

Consider what is predicted to happen on reservoir depressurization in Figure 6. If we imagine a

reservoir fluid at 0.2 mass fraction of gas and 10,000 psia, the asphaltenes are soluble in the oil.

On reservoir depressurization to just below 8,000 psia, asphaltenes begin to come out of solution.

On further depressurization, we reach the bubble point of the oil where the light ends begin to

come out of solution. As the pressure is decreased further, more light ends leave the solution

until at just below 3,000 psia, the oil becomes stable for asphaltenes again. This explains the

field observation that asphaltenes precipitate only in a certain pressure range. Of course, the

19

effect of asphaltene re-dissolution kinetics, which may be significant, is not considered in the

thermodynamic model.

We find similar phase behavior on N2 injection for an oil studied by Jamaluddin , et al.60 In this

system, the parameters for the components in the oil are determined by the same approach as

previously explained in this chapter. One difference is that N2 and CO2 are treated as separate

components because each is present in reasonably high concentration. Also, the parameters for

the asphaltene component are fit to the asphaltene instability pressure prior to N2 injection.

Details of the calculations are presented in reference (61). A comparison of PC-SAFT results

with experimental results is given in Figure 7. The injection of 5, 10, and 20 mole% of N2

strongly increases the asphaltene instability onset. The oil is originally unstable probably due to

its initial high content of CO2 . The difference between the asphaltene onset pressure and the

bubble point pressure (Ponset – Pbbp) increases with the amount of injected N2. The SAFT

predictions closely follow the experimental findings.

For this same system before N2 injection, the pressure / temperature isopleth projection of the

phase diagram has been measured. In Figure 8 we compare the asphaltene onset as a function of

temperature with experimental data.60 PC-SAFT predicts, in qualitative agreement with

experiment, asphaltene instability at low temperature and at high temperature. This type of

phase boundary is commonly seen in mixtures of components with large size asymmetry. At low

temperature, differences in van der Waals interactions and molecular size between the asphaltene

molecules and solvent (crude oil) cause phase separation. At high temperature, the oil becomes a

20

poor solvent and the solution demixes because the light components cause the oil to expand with

increase in temperature.

II.D Effect of Asphaltene Polydispersity on Phase Behavior

Results thus far have shown19, 20, 61 that PC-SAFT can adequately describe asphaltene phase

behavior in a recombined oil under reservoir conditions. In these calculations the asphaltenes

were treated as a single, monodisperse component in oil. Since asphaltenes are, in actuality, a

polydisperse class of compounds with resins as their lowest molecular weight sub-fraction, the

effects of asphaltene polydispersity need to be considered.

Here we examine the effect of asphaltene polydispersity on asphaltene’s thermodynamic phase

behavior in oil. We will show that the lowest molecular weight asphaltenes (including the

resins) can stabilize, via nonpolar interactions, the higher molecular weight asphaltenes. We also

show, via modeling analyses, qualitative differences in the behavior of the various asphaltene

(and resin) sub-fractions. The polydispersity of asphaltenes has been observed in the laboratory,

both in terms of the amount of asphaltenes precipitated with various alkane precipitants and in

the characterization of the precipitated asphaltenes as “hard” or “soft” asphaltenes. Since

deposition tendencies onto pipeline surfaces have often been associated with variations in the

morphology and the composition of the precipitated asphaltene phase, an understanding of the

molecular weight distribution of polydisperse asphaltenes in equilibrium phases should help us

gain a better understanding of the various asphaltene sub-fraction’s deposition tendencies.

21

In this section of the work, polydisperse asphaltenes are represented as four pseudo-components

in SAFT. These pseudo-components are denoted n-Ci-j, representing the asphaltene fraction

insoluble in the n-alkane of carbon number n-Ci, but soluble in the n-alkane n-Cj. Thus, the

pseudo-components are denoted n-C3-5, n-C5-7, n-C7-15, and n-C15+. Hence, n-C15+ asphaltenes

are the asphaltenes fraction insoluble in pentadecane. This assumes, for example, that no n-C7-15

asphaltene will precipitate on addition of pentadecane. We will see that there is some co-

precipitation, leading to an iterative division of pseudo-components. Since we are interested in

the qualitative effects of polydispersity at this point, we consider these pseudo-components

suitable for our purposes. The reader should keep in mind that, traditionally, the n-C3-5 fraction

is called resins and that conventional asphaltene extraction techniques generally identify

asphaltenes as the n-C5+ or n-C7+ insoluble fractions of heavy organics. All asphaltene sub-

fractions in this work are soluble in aromatic solvents (assessed via the PC-SAFT equation of

state).

II.D.1 Selection of PC-SAFT Parameters for Polydisperse Asphaltenes

In experiments performed by Wang,14 the asphaltenes from an oil were separated into various

solubility fractions using excess n-pentane, n-heptane, and n-pentadecane precipitants; these sub-

fractions are called n-C5 insoluble asphaltenes, n-C7 insoluble asphaltenes, and n-C15 insoluble

asphaltenes, respectively. The asphaltene instability onsets with n-alkane titrations were

measured for mixtures of toluene with each isolated asphaltene sub-fraction (at ambient

conditions and with an asphaltene/toluene ratio of 1 g per 100mL toluene). The experimental

asphaltene fractionation and titration data for Lagrave oil from Wang14 were used to fit

parameters for each asphaltene pseudocomponent.

22

Using this model, we could then study how polydispersity and resins affect the stability of

asphaltenes. Although each sub-fraction, including the resins, is polydisperse, we model

asphaltenes as either 3 or 4 pseudo-components (depending on whether the resin fraction, which

is the n-C3-5 asphaltene sub-fraction, was included): an n-C15+ sub-fraction, an n-C7-15 sub-

fraction, an n-C5-7 sub-fraction, and an n-C3-5 sub-fraction. The method used to fit the SAFT

parameters for each asphaltene pseudo-component was similar in principle to the monodisperse

SAFT asphaltene characterization procedure discussed above and given in Ting, et al.19, 20.

Succinctly:

(1) SAFT parameters were fit for the n-C15+ asphaltene sub-fraction to reproduce the

experimental data on the minimum volume fraction of alkane precipitant needed to

induce asphaltene instability (φvppt) for mixtures of n-C15 insoluble asphaltene,

toluene, and various n-alkanes (see Figure 9).

(2) The asphaltene made from the combination of n-C15+ and n-C7-15 sub-fractions were

assumed to represent the n-C7 insoluble asphaltenes; a second set of SAFT parameters

were fit for the n-C7-15 sub-fraction to reproduce (together with the previously fitted

n-C15+ sub-fraction) the experimental φvppt data for a mixture of n-C7 insoluble

asphaltene, toluene, and n-alkane as shown in Figure 9.

(3) A third set of SAFT parameters were fit for the n-C5-7 sub-fraction so that the

combination of the n-C15+ (fitted in Step (1)), n-C7-15 (fitted in Step (2)), and n-C5-7

23

sub-fractions represented the n-C5 insoluble asphaltenes and reproduced the

experimental φvppt data for a mixture of n-C5 insoluble asphaltene, toluene, and n-

alkanes as shown in Figure 9.

(4) Due to lack of precipitation data, the PC-SAFT parameters for the resin (n-C3-5) sub-

fraction were obtained by decreasing the PC-SAFT parameters (m and ε/k) of the n-

C5-7 asphaltene sub-fraction until a set of PC-SAFT parameter was obtained that

would make the resins insoluble in propane (tested at 10 bars) and soluble in n-

pentane.

Because there is insufficient data to uniquely fit all of the model parameters for polydisperse

asphaltenes, certain approximations and relationships have to be made:

• The molecular weights of all PC-SAFT asphaltene sub-fractions were set to be linearly

dependent on chain length, m. This was done because the experimental molecular weight

of each asphaltene sub-fraction is not known and because the SAFT chain length is

roughly linearly dependent on molecular weight for polynuclear aromatics. The constant

of proportionality (MW=m/0.0216) used in this work was set to give the n-C15+

asphaltene sub-fraction a molecular weight of 2,500.

• The segment diameters of all asphaltene sub-fractions were set to 4 Angstroms. This is

an average value of the segment diameters for most polynuclear aromatics and

polynuclear aromatics derivatives.

24

A comparison of the equation-of-state fitted and the experimental φvppt data is shown in Figure 9,

with the fitted PC-SAFT asphaltene parameters listed in Table 4. As seen in the figure, the

agreement between PC-SAFT calculated and measured φvppt is qualitative. PC-SAFT is able to

describe the change in magnitude (and to a lesser extent, the curvature) of the φvppt vs. n-alkane

carbon number curve between n-C15 insoluble and n-C5 insoluble asphaltenes. For the

precipitation onsets with n-C7 and n-C5 extracted asphaltenes, PC-SAFT underestimates the

experimental φvppt data in cases where larger n-alkanes (undecane and higher) are used to induce

asphaltene precipitation. As seen in the figure, PC-SAFT predicts a maximum in the volume

fraction of precipitant at a carbon number of about 9, in agreement with previous experimental

observations.

A plot of the SAFT parameters ε/k vs. m for the various SAFT asphaltene sub-fractions and resin

(the n-C3-5 sub-fraction) shows a well-defined trend between ε/k and m (Figure 10). The

asphaltenes precipitated by the lower molecular weight n-alkanes tend to be smaller in size and

have lower segment energy. It is also interesting to note that despite having the same segment

size, the segment energy of the “heavier” asphaltene sub-fractions seem to be larger. The

asphaltene sub-fractions have a segment energy close to that for naphthalene and alpha-

methylnaphthalene (Table 1 and Figure 2); however, the chain lengths, m, of the asphaltenes are

much larger. Finally, the chain length (m) for the resin sub-fraction seems to be much smaller

than the other asphaltene sub-fractions. Equation of state parameters for the asphaltene and resin

fractions are given in Table 4.

25

II.D.2 Effects of Asphaltene Polydispersity and Resin Addition

To investigate the roles resins and asphaltene polydispersity play on asphaltene phase behavior

in oil, we will compare the phase behavior of four model oil mixtures containin monodisperse or

polydisperse asphaltenes. In all model systems, toluene is the model oil and the asphaltenes have

a fixed concentration of 7.5 g asphaltenes to 100 mL of toluene. The properties and

asphaltene/resin contents of these systems are listed in Table 5 and are discussed in the following

paragraph. For these model system investigations, the focus is on the qualitative

trends/behaviors and all binary interaction parameters between all species are set to zero

(following Ting, et al. 19, 20).

The biggest difference between the various systems in Table 5 is that asphaltenes in Systems 1

and 2 are monodisperse, while the asphaltenes in Systems 3 and 4 are polydisperse. More

specifically, the asphaltene used in Model System 1 is monodisperse and was fit to experimental

φvppt data for the n-C15 insoluble Lagrave asphaltenes14 (also called the n-C15+ asphaltene fraction

in this work). The asphaltene used in Model System 2 is monodisperse and was fit to

experimental φvppt data for the n-C5 insoluble Lagrave asphaltenes. The asphaltene used in

Model System 3 is polydisperse and the SAFT parameters for each asphaltene sub-fraction were

fit to the experimental φvppt data of the fractionated asphaltenes. Model System 4 is similar to

Model System 3 with the exception that 10g of resin per 100mL toluene (approximately 1-2

moles resin/100moles toluene) is added to the system. The amount of resin added (10g resin per

100mL toluene) is arbitrary.

26

The effects of n-alkane addition on the amount of asphaltenes precipitated (at 20°C and 1 bar) for

the four model oil mixtures are shown in Figures 11 and 12. For systems containing

monodisperse asphaltenes (Figure 11), the change in asphaltene solubility is dramatic:

asphaltenes go from completely soluble to almost completely insoluble in the model oil when the

volume fraction of the n-alkane precipitant is increased slightly past the asphaltene instability

onset point. As expected, the lower molecular weight asphaltenes (the monodisperse asphaltene

fit to the n-C5 insoluble asphaltenes φvppt data) are more soluble than the higher molecular weight

asphaltene (the monodisperse, n-C15 insoluble asphaltene) in terms of the amount of precipitant

needed to induce asphaltene instability. When sufficiently large amount of n-alkanes are added

to the model oil, all asphaltenes will precipitate.

A large change in the amount of precipitated asphaltenes vs. precipitant volume fraction can be

seen when the effect of asphaltene polydispersity (and resin addition) is taken into consideration

(Figure 12). By treating asphaltene as a polydisperse specie, the amount of asphaltenes

precipitated increases much more gradually with precipitant addition. A significant amount of

asphaltenes will stay in solution even at high precipitant volume fractions, and more asphaltenes

can be precipitated using lower molecular weight n-alkanes. It is interesting to note that when n-

C15 is used as the precipitant, SAFT predicts the existence of a solubility minimum around

φvppt=0.9.

A comparison of the SAFT-predicted behavior of polydisperse asphaltenes with and without

resins show that the presence of resins will increase the amount of precipitant needed to induce

the onset of asphaltene instability (Figure 12). Furthermore, at lower precipitant volume

27

fractions in the oil, the amount of asphaltenes that will precipitate is less when resins are present.

Even though only dispersion interactions are considered in these PC-SAFT models, the lower

molecular weight asphaltenes and especially resins will stabilize the heavier asphaltenes in the

oil. It can be seen in Figure 12 that the effects of resins on asphaltene phase behavior in the oil

become less pronounced as the oil is diluted more with precipitants.

A plot of the mass distribution of the asphaltene sub-fractions as a function of precipitant volume

fraction is shown in Figure 13. As seen in the figure, near the initial asphaltene instability onset,

the precipitated phase is composed mostly of the heaviest asphaltene fractions (in this case, the

n-C15+ sub-fraction). As the amount of precipitant is increased further, more and more lower

molecular weight asphaltenes will precipitate.

III. Summary & Conclusions

The effect of pressure, temperature, and composition on the phase behavior of asphaltenes in

crude oil systems can be explained in terms of van der Waals interactions between molecules

using the PC-SAFT equation of state. A method was introduced to characterize the recombined

oil including asphaltenes as a six-component mixture using the PC-SAFT equation of state. The

real components and pseudo-components were chosen based on saturates-aromatics-resins-

asphaltenes (SARA) fractionation, gas chromatography, and gas-oil ratio information. Equation

of state parameters for each component (except asphaltenes) were determined from generalized

correlations in terms of molecular weight. Binary interaction parameters were fit to vapor liquid

equilibria data. Asphaltene parameters were determined by modeling asphaltene precipitation

upon titration of oil with n-alkanes at ambient conditions. PC-SAFT was found to accurately

28

predict the density, the bubble point curve and the asphaltene instability region for the

recombined oil over a range of temperatures, pressures, and compositions.

The PC-SAFT equation of state has proved useful in explaining laboratory and field observations

of asphaltene phase behavior. For example, when asphaltene precipitates near the bubble point,

the model predicts asphaltene stability at high pressure and at pressures well below the bubble

point. On titration of the oil with various n-alkanes, we show that the volume fraction of

precipitant required for asphaltene precipitation as a function of carbon number of the alkane

precipitant reaches a maximum at about nonane. The model also predicts that asphaltenes can

become unstable as temperature is decreased or increased depending on the composition of the

oil and the temperature. The model has also proved to be predictive in describing the effect of

gas injection (nitrogen and methane) and of alpha methyl naphthalene on asphaltene stability.

We have investigated the effect of polydispersity and resins on asphaltene phase behavior for a

model system. We find that resins delay the onset of asphaltene precipitation, but the total

amount of asphaltene precipitated is unaffected by the resins. SAFT calculations show that the

lower molecular weight asphaltenes and resins play a large role in stabilizing higher molecular

weight asphaltenes in oil. Resin’s stabilizing effects on polydisperse asphaltene is greatest in the

region of incipient asphaltene instability; when sufficiently large amounts of n-alkane

precipitants are added, similar amounts of asphaltenes would precipitate regardless of the

presence of resins in the oil. An analysis of the mass distribution of the asphaltene sub-fractions

in the precipitated phase shows that the largest asphaltenes will precipitate first, followed by the

29

precipitation of smaller asphaltenes upon further oil dilution. In our study, the heaviest

asphaltene sub-fraction precipitated first.

The amount and type of data presently available for onset of asphaltene precipitation can be

adequately modeled without including the effect of polar groups or association. Although

extensions of the PC-SAFT equation of state can account for these effects, presently available

data is insufficient to justify the additional parameters needed to include these effects.

Acknowledgements

We gratefully acknowledge the Department of Energy, DeepStar, ChevronTexaco, DB

Robinson, and the Consortium of Processes in Porous Media at Rice University for their

financial support. We also thank Jeff Creek and Jill Buckley for many helpful discussions.

References 1 W.G. Chapman, G. Jackson, and K.E. Gubbins, Mol. Phys. 65, 1057-1079, (1988). 2 W.G. Chapman, K.E. Gubbins, G. Jackson, and M. Radosz, Fluid Phase Equil. 52, 31-38,

(1989). 3 W.G. Chapman, K.E. Gubbins, G. Jackson, and M. Radosz, Ind. Eng. Chem. Res. 29,

1709-1721, (1990). 4 J. Gross,. and G. Sadowski, Ind. Eng. Chem. Res. 40, 1244-1260, (2001). 5 I.A. Wiehe and K.S. Liang, Fluid Phase Equil. 117, 201-221, (1996). 6 C.E. Haskett and M. Tartera, J. Pet. Tech. 17, 387-391, (1965). 7 R.N. Tuttle, J. Pet. Tech. 35, 1192-1196, (1983). 8 S. Adialalis, Investigation of physical and chemical criteria as related to the prevention

of asphalt deposition in oil well tubings. Ph.D. Thesis, Imperial College, 1982 9 A.J.L. Chavez and M.A. Lory, Revista del Instituto Mexicano del Petroleo 23, 1, (1991). 10 J. Escobedo and G.A. Mansoori, in SPE Annual Technical Conference and Exhibition

(Dallas, TX, 1995). 11 C. von Albrecht, W.M. Salathiel, and D.E. Nierode, CIM Special Volume , 55-62, (1977).

30

12 J. Wu, Molecular Thermodynamics of Some Highly Asymmetric Liquid Mixtures, Ph.D. Thesis, University of California at Berkeley, 1998.

13 J.M. Prausnitz, R.N. Lichtenthaler, and E.G. de Azevedo, Molecular Thermodynamics of Fluid-Phase Equilibria, 2nd ed. P T R Prentice-Hall, Inc., 1986.

14 J.X. Wang, Predicting Asphaltene Flocculation in Crude Oils, Ph.D. Thesis, New Mexico Institute of Mining & Technology, 2000.

15 A. Hirschberg, L.N.J. deJong, B.A. Schipper, and J.G. Meijer, J. G., Soc. Pet. Eng. J. 24, 283-293, (1984).

16 K. Akbarzadeh, S. Ayatollahi, M. Moshfeghian, H. Alboudwarej., W.Y. Svrcek, and H.W. Yarranton, in Amer. Inst. Chem. Eng. Spring National Meeting (New Orleans, LA, 2002), pp. 218-224.

17 R. Cimino, S. Correra, and P.A. Sacomani, in Soc. Pet. Eng. Intl. Symp. on Oilfield Chem. (San Antonio, TX, 1995).

18 L.X. Nghiem, M.S. Hassam, R. Nutakki, and A.E.D. George, in Soc. Pet. Eng. Annual Tech. Conf. and Exhib. (Houston, TX, 1993), pp. 375-384.

19 P.D. Ting, G.J Hirasaki, and W.G. Chapman, Pet. Sci. Tech. 21, 647-661, (2003). 20 P.D. Ting, Thermodynamic Stability and Phase Behavior of Asphaltenes in Oil and of

other highly Asymmetric Mixtures, Ph.D. Thesis, Rice University, 2003. 21 F. Chung, P. Sarathi, and R. Jones, (National Institute for Petroleum and Energy

Research, Bartlesville, 1991). 22 N.E. Burke, R.E. Hobbs, and S.F. Kashou, J. Pet. Tech. 42, 1440-1446 (1990). 23 J.X. Wang, J.S. Buckley, N.A. Burke, and J.L. Creek, paper 15254 presented at the 2003

SPE Offshore Technology Conference, Houston, May 5-7; J.X. Wang and J.S. Buckley, Energy & Fuels, 15, 1004-1012, (2001).

24 L.X. Nghiem and D.A. Coombe, Soc. Pet. Eng. J. 2 (1997). 25 P.D. Ting, P.C. Joyce, P.K. Jog, W.G. Chapman, and M.C. Thies, Fluid Phase Equil.,

(2002). 26 K.J. Leontaritis. and G.A. Mansoori, in Soc. Pet. Eng. Intl. Symp. on Oilfield Chem. (San

Antonio, TX, 1987). 27 A.I. Victorov and A. Firoozabadi, Amer. Inst. Chem. Eng. J. 42, 1753-1764, (1996). 28 J. Wu, J.M. Prausnitz, and A. Firoozabadi, Amer. Inst. Chem. Eng. J. 44, 1188-1199,

(1998). 29 J. Wu, J.M. Prausnitz, and A. Firoozabadi, Amer. Inst. Chem. Eng. J. 46, 197-209,

(2000). 30 Bungert, B. Komplexe Phasengleichgewichte von Polymerlösungen. Dissertation,

Technische Universität Berlin, Germany, 1998; J. Gross and G. Sadowski, Ind. Eng. Chem. Res., 41, 1084-1093, (2002).

31 I.A. Wiehe, Fuel Sci. Tech. Intl. 14, 289-312, (1996). 32 E.W. Moore, C.W. Crowe, and A.R. Hendrickson, J. Pet. Tech. 17, 1023-1028, (1965). 33 C.L. Chang and H.S. Fogler, in Soc. Pet. Eng. Intl. Symp. on Oilfield Chem. (New

Orleans, LA, 1993). 34 H. Groenzin and O.C. Mullins, J. Phys. Chem. A, 103, 11237-11245, (1999). 35 I.G. Economou, Ind. Eng. Chem. Res. 41, 953-962, (2002). 36 E.A. Muller. and K.E. Gubbins, Ind. Eng. Chem. Res. 40, 2193-2212, (2001). 37 P.K. Jog, W.G. Chapman, S.K. Gupta, and R.D. Swindoll, Ind. Eng. Chem. Res., 41, 887-

891, (2002).

31

38 M.S. Wertheim, J. Stat. Phys. 35, 19, (1984). 39 M.S. Wertheim, J. Stat. Phys. 35, 35, (1984). 40 M.S. Wertheim, J. Stat. Phys. 42, 459, (1986). 41 M.S. Wertheim, J. Stat. Phys. 42, 477, (1986). 42 W.G. Chapman, J. Chem. Phys., 93, 4299-4304, (1990). 43 T. Kraska and K.E. Gubbins, Ind. Eng. Chem. Res., 35, 4727-4737, (1996); Ind. Eng.

Chem. Res., 35, 4738-4746, (1996) 44 A. GilVillegas, A. Galindo, P.J. Whitehead, S.J. Mills, G. Jackson, A.N. Burgess, J.

Chem. Phys., 106, 4168-4186 (1997). 45 S.H. Huang and M. Radosz, Ind. Eng. Chem. Res. 29, 2284-2294, (1990). 46 DIPPR, (AIChE, 2003). 47 P.D. Ting, J. Wang, J.S. Buckley, J. Creek, A. Hamammi, G.J. Hirasaki, and W.G.

Chapman, In Preparation, (2005). 48 A.F. Kidnay, R.C. Miller, W.R. Parrish, and M.J. Hiza, Cryogenics 15, 531-540, (1975). 49 H.H, Reamer, B.H. Sage, and W.N. Lacey, Ind. Eng. Chem. 42, 534, (1950). 50 L. Grauso, A. Fredenslund, and J. Mollerup, Fluid Phase Equil. 1, 13-26, (1977). 51 H.H. Reamer, R.H. Olds, B.H. Sage, and W.N. Lacey, Ind. Eng. Chem. 34, 1526-1531,

(1942). 52 M. Elbishlawi and J.R. Spencer, Ind. Eng. Chem. 43, 1811-1815, (1951). 53 A. Azarnoosh and J.J. McKetta, J. Chem. Eng. Data 8, 494-496, (1963). 54 P.C. Joyce, Vapor-Liquid Equilibria for Long-Chain Hydrocarbons in Supercritical

Alkane Solvents. Ph.D. Thesis, Clemson University, 1999; P.C. Joyce, J. Gordon, and M.C. Thies, J. Chem. Eng. Data 45, 424-427, (2000); P.C. Joyce and M.C. Thies, J. Chem. Eng. Data 43, 819-822, (1998).

55 P. Miller and B.F. Dodge, Ind. Eng. Chem. 32, 434-438, (1940). 56 J.W. Glanville, B.H. Sage, and W.N. Lacey, Ind. Eng. Chem. 42, 508, (1950). 57 U. Messow and I. Engel, Z. Phys. Chem. 258, 798, (1977). 58 J.S. Buckley and J. Wang, Private Communication, (2003). 59 J.S. Buckley, G.J. Hirasaki, Y. Liu, S. von Drasek, J. Wang, and B.S. Gill, Pet. Sci. Tech.

16, 251-285, (1998). 60. A.K.M. Jamaluddin, N. Joshi, F. Iwere, and O. Gurnipar, SPE 74393, (2001). 61. D.L. Gonzalez, P.D. Ting, G.J. Hirasaki, and W.G. Chapman, Energy & Fuels, 19, 1230-

1234 (2005).

32

Tables Table 1. PC-SAFT pure component parameters: m is the number of segments that make up a

molecule, σ is the segment diameter in angstroms, ε/k is the interaction energy between a pair of segments in K. For substances fitted by Ting,20 the AAPDsa and temperature range of the experimental data are given.46 All other PC-SAFT parameters are from Gross and Sadowski.4

AAPDa

Substance MW (g/mol) m σ

(A) ε/k (K)

T range (K) Psat ρliq

n-alkanes methane 16.04 1.0000 3.7039 150.03

ethane 30.07 1.6069 3.5206 191.42

propane 44.09 2.0020 3.6184 208.11

butane 58.12 2.3316 3.7086 222.88

pentane 72.15 2.6896 3.7729 231.20

hexane 86.18 3.0576 3.7983 236.77

heptane 100.20 3.4831 3.8049 238.40

octane 114.23 3.8176 3.8373 242.78

nonane 128.25 4.2079 3.8448 244.51

decane 142.29 4.6627 3.8384 243.87

undecane 156.31 4.9082 3.8893 248.82

dodecane 170.34 5.3060 3.8959 249.21

tridecane 184.37 5.6877 3.9143 249.78

tetradecane 198.39 5.9002 3.9396 254.21

pentadecane 212.42 6.2855 3.9531 254.14

hexadecane 226.45 6.6485 3.9552 254.70

eicosane 282.55 7.9849 3.9869 257.75

Cycloalkanes cyclopentane 70.13 2.3655 3.7114 265.83

methyl-cyclopentane 84.16 2.6130 3.8253 265.12

ethyl-cyclopentane 98.18 2.9062 3.8873 270.50

cyclohexane 84.15 2.5303 3.8499 278.11

methyl-cyclohexane 98.18 2.6637 3.9993 282.33

ethyl-cyclohexane 112.22 2.8256 4.1039 294.04

cyclopheptane 98.19 2.6870 3.9352 296.15 300-570 0.44 0.21

cyclooctane 112.22 2.9222 4.0028 304.67 300-620 0.71 1.06

polynuclear aromatics benzene 78.11 2.4653 3.6478 287.35

naphthalene 128.17 3.0915 3.8333 348.40 373-633 1.45 0.77

anthracene 178.23 3.5291 4.0922 402.13 500-830 1.81 1.57

phenanthrene 178.23 3.4890 4.1053 403.06 330-780 1.51 1.39

naphthacene 228.29 4.6432 3.8942 407.60 660-940 0.88 4.23

chrysene 228.29 5.1201 3.8400 385.73 580-940 0.81 3.61

pyrene 202.26 3.6847 4.1151 427.35 450-850 1.37 3.76

aromatic & polynuclear aromatics derivatives

33

toluene 92.14 2.8149 3.7169 285.69

ethylbenzene 106.17 3.0799 3.7974 287.35

propylbenzene 120.19 3.3438 3.8438 288.13

butylbenzene 134.22 3.7662 3.8727 283.07

tetralin 132.21 3.3131 3.8750 325.07

biphenyl 154.21 3.8877 3.8151 327.42

1-methylnaphthalene 142.20 3.4064 3.8961 345.71 393-673 0.29 0.50

1-phenylnaphthalene 204.27 4.7634 3.8582 336.53 330-780 1.51 1.39

m-terphenyl 228.29 5.6273 3.7967 329.18 420-840 1.52 0.26

Gases nitrogen 28.01 1.2053 3.3130 90.96

carbon dioxide 44.01 2.0729 2.7852 169.21

carbon disulfide 76.14 1.6919 3.6172 334.82

a Average absolute percent deviation = %1001�

−ndp alExperiment

alExperimentCalculatedndp

Table 2. Fractionation of the recombined oil with GOR = 152 m3/m3 into a 6 component

mixture.

Component Mole Fraction Average

MW m σ��(A) ε/k(K)

Methane 0.3300 16 1.0000 3.7039 150.03

Nitrogen & Carbon Dioxide 0.0985 28 1.2053 3.3130 90.96

Light Alkanes 0.2234 47 1.9854 3.6917 206.12

Saturates 0.2445 255 7.3765 3.9635 257.12

Aromatics & Resins 0.1023 212 5.3351 3.7637 323.70

Asphaltenes 0.0013 1700* 29.5000 4.3000 395.00 * molecular weight of pre-aggregated asphaltene

34

Table 3. Binary interaction parameters used in this study.

Component methane N2 + CO2 Light

Alkanes Saturates Aromatics & Resins

Asphaltenes

methane 0 0a 0.01b 0.03d 0.03e 0.03k N2 + CO2 0a 0 0.07c 0.01f 0.13h 0.13 k

Light Alkanes 0.01b 0.07c 0 0.006g 0.02i 0.02k Saturates 0.03d 0.01f 0.006g 0 -0.01j 0

Aromatics & Resins 0.03

e 0.13h 0.02i -0.01j 0 0

Asphaltenes 0.03 k 0.13 k 0 0 0 0

a based on methane-N2 binary data of Kidnay, et al.48 b based on methane-propane binary data of Reamer, et al.49 c based on N2-propane binary data of Grauso, et al.50 d based on methane-decane data of Reamer, et al.51 e based on methane-toluene binary data reported in this work and from Elbishlawi and Spencer52 f based on N2-decane binary data of Azarnoosh and McKetta53 g based on propane-hexadecane binary data of Joyce54 h based on N2-benzene binary data of Miller and Dodge55 i based on propane-benzene binary data of Glanville, et al.56 j based on toluene-dodecane and toluene-hexadecane binary data of Messow and Engel57 k the kijs’ between asphaltenes and separator gas sub-fractions are assumed to be the same as the kijs’ between aromatics+resins and separator gas sub-fractions. Table 4. PC-SAFT parameters for the various asphaltene sub-fractions (including resins)

SAFT Parameters Asphaltene sub-fraction MW m σ (A) ε/k (K) δ (MPa0.5) ρ (g/cm3)

n-C15+ 2500* 54 4.00 350.5 22.17 1.150

n-C7-15 1852* 40 4.00 340.0 21.52 1.137 n-C5-7 1806* 39 4.00 335.0 21.25 1.133

Resin 556 12 4.00 330.0 20.41 1.103 Monodisperse – n-C5 2080* 46 4.00 350.5 22.13 1.120

* molecular weight of pre-aggregated asphaltene

35

Tables 5. Description of four representative model oils tested to study the effects of

asphaltene polydispersity and resin addition. All model oils contained 7.5 g of asphaltene to 100 mL of toluene (20°C, 1 bar).

Model System Included Asphaltene fractions

System 1 Monodisperse – n-C15+ subfraction only

System 2 Monodisperse – n-C5 System 3 n-C15+:n-C7-15: n-C5-7 with mass ratios of 4.5 : 2.0 : 3.5

System 4 n-C15+:n-C7-15: n-C5-7 with mass ratios of 4.5 : 2.0 : 3.5 and resins 10 g/100 mL toluene (20°C, 1 bar)

36

Figure Captions Figure 1. Contributions to the SAFT equation of state for an associating polyatomic fluid. Figure 2. Plots of PC-SAFT pure component parameters as a function of molecular weight for

saturates and polynuclear aromatics. Figure 3. Representation of a recombined oil with six pseudo-components. Figure 4. Refractive index at the point of asphaltene precipitation (PRI) from an equi-volume

mixture of stock tank oil and alpha-methyl naphthalene on titration with n-alkane. The curve shows results from the PC-SAFT equation of state and the data is from Wang and Buckley.58

Figure 5. Experimental47 and PC-SAFT-predicted (lines) one-phase recombined oil densities with two different gas-oil ratios at 71.1°C.

Figure 6. PC-SAFT-predicted and measured19, 47 asphaltene instability onset and mixture bubble points for the recombined oil at 71.1°C.

Figure 7. PC-SAFT-predicted and measured60 asphaltene instability onset and mixture bubble points as a function of nitrogen concentration in a recombined oil at 296°F.

Figure 8. The temperature dependence of the asphaltene instability curve and bubble curve predicted by PC-SAFT and experimental measurements60 for the reservoir fluid prior to nitrogen injection.

Figure 9. Comparison of PC-SAFT and measured precipitant volume fraction at asphaltene instability onset for asphaltene-toluene-n-alkane mixtures at 20°C and 1 bar. The asphaltene/toluene ratio is 1g/100mL. Experimental data are from Wang14

Figure 10. Plot of ε/k vs. m for the various PC-SAFT asphaltene sub-fractions and resin. Figure 11. Solubility of monodisperse asphaltenes in model oil (7.5g asphaltene/100mL

toluene) mixed with n-alkanes at 20°C and 1 bar. Figure 12. Solubility of polydisperse asphaltenes (with or without resins) in model oil (7.5g

total asphaltene/100mL toluene) mixed with n-alkanes at 20°C and 1 bar. Figure 13. Normalized distribution of the asphaltene sub-fractions in the precipitated phase as

a function of volume fraction precipitant in the model oil mixtures.

37

Segments Molecules Associating Molecules

assocchainsegment AAAA ∆+∆+=

Figure 1. Contributions to the SAFT equation of state for an associating polyatomic fluid.

Figure 2. Plots of PC-SAFT pure component parameters as a function of molecular weight for

saturates and polynuclear aromatics.

38

Figure 3. Representation of a recombined oil with six pseudo-components.

Figure 4. Refractive index at the point of asphaltene precipitation (PRI) from an equi-volume

mixture of stock tank oil and alpha-methyl naphthalene on titration with n-alkane. The curve shows results from the PC-SAFT equation of state and the data is from Wang and Buckley.58

Saturates 24%

Light Alkanes 22% molN2+CO2 9.9%

Methane 33% mol Aromatics &

Resins 10% molAsphaltenes 0.001% mol

Methane MWn = 16 N2+CO2 MWn = 28Light Alkanes MWn = 47Saturates MWn = 255Aromatics & Resins MWn = 222Asphaltene “aggregate” MWn = 1700

Gas Density (3495 psi, 67.7F) = 0.377 g/ccOil Density (14.9 psi, 67.7F) = 0.857 g/cc Oil Density (14.9 psi, 130.7F) = 0.848 g/cc GOR = 152 m3/m3Psat (160F) = 4250 psia

39

Figure 5. Experimental47 and PC-SAFT-predicted (lines) one-phase recombined oil densities

with two different gas-oil ratios at 71.1°C.

40

Figure 6. PC-SAFT-predicted and measured19, 47 asphaltene instability onset and mixture

bubble points for the recombined oil at 71.1°C.

41

0

2000

4000

6000

8000

10000

12000

0 5 10 15 20 25

N2 injected, mole%

Pre

ssur

e (p

sia)

VLE region

Stable region

Unstable region

Reservoir 'A' fluid 296 F

Figure 7. PC-SAFT-predicted and measured60 asphaltene instability onset and mixture bubble

points as a function of nitrogen concentration in a recombined oil at 296°F.

42

0

1000

2000

3000

4000

5000

6000

0 100 200 300 400 500 600

Temperature (oF)

Pre

ssur

e (p

sia)

Reservoir 'A' fluid

Bubble point curve

Unstable region

Stable region

Figure 8. The temperature dependence of the asphaltene instability curve and bubble curve

predicted by PC-SAFT and experimental measurements60 for the reservoir fluid prior to nitrogen injection.

43

0.30

0.40

0.50

0.60

0.70

5 7 9 11 13 15

n-alkane carbon number

pre

cip

itan

t vo

lum

e fr

actio

n n-C15 asph (expt)n-C7 asph (expt)

n-C5 asph (expt)

SAFT

Figure 9. Comparison of PC-SAFT and measured precipitant volume fraction at asphaltene

instability onset for asphaltene-toluene-n-alkane mixtures at 20°C and 1 bar. The asphaltene/toluene ratio is 1g/100mL. Experimental data are from Wang14

320

325

330

335

340

345

350

355

360

0 10 20 30 40 50 60

m

/k (K

)

n-C15+ sub-frac

n-C5-7 sub-frac

n-C7-15 sub-frac

resins

Figure 10. Plot of ε/k vs. m for the various PC-SAFT asphaltene sub-fractions and resin.

44

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1precipitant volume fraction

amou

nt a

sph

alte

ne

prec

ipita

ted

ove

r to

tal a

mou

nt a

sph

alte

ne

(mas

s)

n-C15 precipitantModel System 1

n-C5 precipitantModel System 1

n-C15 precipitantModel System 2

n-C5 precipitantModel System 2

Unstable Region

Stable Region

Figure 11. Solubility of monodisperse asphaltenes in model oil (7.5g asphaltene/100mL

toluene) mixed with n-alkanes at 20°C and 1 bar.

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

precipitant volume fraction

amou

nt a

spha

ltene

pre

cipi

tate

d ov

er to

tal a

mou

nt a

spha

lten

e (m

ass)

precipitant = n-C15polydisperse asphaltenes

Model System 3

precipitant = n-C5polydisperse asphaltenes

Model System 3

precipitant = n-C5polydisperse asphaltenes + resins

Model System 4

precipitant = n-C15polydisperse asphaltenes + resins

Model System 4

UnstableRegion

StableRegion

Figure 12. Solubility of polydisperse asphaltenes (with or without resins) in model oil (7.5g

total asphaltene/100mL toluene) mixed with n-alkanes at 20°C and 1 bar.

45

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

precipitant volume fraction

amou

nt a

spha

ltene

pre

cipi

tate

dov

er to

tal a

mou

nt a

spha

ltene

(m

ass)

precipitant = n-C15dashed lines = polydisperse asph (Model System 3)solid lines = polydisperse asph + resins (Model System 4)

n-C15+ sub-fraction

n-C7-15 sub-fraction

n-C5-7 sub-fraction

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

precipitant volume fraction

amou

nt a

spha

ltene

pre

cipi

tate

dov

er to

tal a

mou

nt a

sph

alte

ne

(mas

s)

precipitant = n-C5dashed lines = polydisperse asph (Model System 3)solid lines = polydisperse asph + resins (Model System 4)

n-C15+ sub-fraction

n-C7-15 sub-fraction

n-C5-7 sub-fraction

Figure 13. Normalized distribution of the asphaltene sub-fractions in the precipitated phase as

a function of volume fraction precipitant in the model oil mixtures.